Tinyscript steganography tool based on base32/64 padding

README.md

Paddinganograph

This Tinyscript-based tool allows to unhide data hidden in base32/base64 strings. It can take a PNG or JPG in input to retrieve an EXIF value as the input data.

This can be installed using:

$ pip install tinyscript
$ tsm install paddinganograph

$ paddinganograph -e base64 -f Comment -s . < test.jpg
[...]
$ paddinganograph -e base64 -f Comment -s . < test.jpg | paddinganograph -e base32
[...]

This tool is especially useful in the use cases hereafter.

Retrieve hidden data from an image using Base32/64 padding

Select the "Comment" field, split its value with the separator "." and extract hidden data:

$ paddinganograph -s "." -f "Comment" < image.jpg
[...]

paddinganograph.py

#!/usr/bin/python3
# -*- coding: UTF-8 -*-
import math
from tinyscript import *


__author__    = "Alexandre D'Hondt"
__version__   = "1.6"
__copyright__ = ("A. D'Hondt", 2019)
__license__   = "gpl-3.0"
__reference__ = "https://inshallhack.org/paddinganography/"
__examples__  = ["-s . -f \"Comment\" < image.jpg > base32.enc", "-e base32 < base32.enc"]
__doc__       = """
*Paddinganograph* allows to unhide data hidden in base32/base64 strings. It can take a PNG or JPG in input to retrieve an EXIF value as the input data.
"""


SCRIPTNAME_FORMAT = "none"

DEF_BASE64 = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"
DEF_BASE32 = "ABCDEFGHIJKLMNOPQRSTUVWXYZ234567"


def exif(raw_data, key):
    fp = ts.TempPath().tempfile()
    with fp.open('wb') as f:
        f.write(raw_data)
    logger.debug("Getting EXIF data...")
    exif = subprocess.check_output(["exiftool", str(fp)])
    fp.remove()
    exif = codext.decode(exif, "utf-8")
    exif = {l.split(':', 1)[0].strip(): l.split(':', 1)[1].strip() for l in exif.split('\n') if l.strip() != ""}
    return exif if not key else exif[key]


def unhide(encoded, encoding="base64", charset=None, sep=".", pad="=", n_pad=8):
    try:
        charset = (charset or globals()["DEF_{}".format(encoding.upper())]).strip()
    except NameError:
        raise ValueError("Bad encoding")
    logger.debug("Unhidding data...")
    bits = b("")
    for token in b(encoded).split(b(sep)):
        bits += unhide_bits(token.strip(), charset, pad, n_pad) or b("")
    return b("").join(b(ts.bin2str(ensure_str(bits[i:i+8]))) for i in range(0, len(bits), 8))


def unhide_bits(encoded, charset, pad="=", n_pad=8):
    def __gcd(a,b):
        while b > 0:
            a, b = b, a % b
        return a
    padding = encoded.count(b(pad))
    n_repr = int(math.ceil(math.log(len(charset), 2)))
    w_len  = n_repr * n_pad / __gcd(n_repr, n_pad)
    n_char = int(math.ceil(float(w_len) / n_repr))
    if encoded == "" or len(encoded) % n_char != 0 or padding == 0:
        return
    unused = {n: int(w_len - n * n_repr) % n_pad for n in range(n_char)}
    b_val  = bin(b(charset).index(encoded.rstrip(b(pad))[-1]))[2:].zfill(n_repr)
    return b(b_val[-unused[padding]:])


if __name__ == '__main__':
    parser.add_argument("-c", "--charset", help="characters set")
    parser.add_argument("-e", "--encoding", choices=["base32", "base64"], default="base64", help="character encoding")
    parser.add_argument("-f", "--exif-field", dest="exif", default="Comment", help="EXIF metadata field to be selected")
    parser.add_argument("-p", "--padding-char", default="=", help="padding character")
    parser.add_argument("-s", "--separator", default="\n", help="base-encoded token separator")
    initialize()
    data = b("").join(l for l in ts.stdin_pipe())
    if data.startswith(b("\x89PNG")) or data.startswith(b("\xff\xd8\xff\xe0")):
        try:
            data = exif(data, args.exif)
        except KeyError:
            logger.error("No EXIF field '%s'" % args.exif)
            sys.exit(1)
    print(ensure_str(unhide(data, args.encoding, args.charset, args.separator)))

@roumy OK, I figured it out ; the offending line is :

return b("").join(ts.bin2str(bits[i:i+8]) for i in range(0, len(bits), 8))

>>> from tinyscript import *
>>> ts.bin2str("01010101")
'U'
>>> ts.bin2str(b"01010101")
Traceback (most recent call last):
  File "/usr/lib/python3.8/idlelib/run.py", line 559, in runcode
    exec(code, self.locals)
  File "<pyshell#10>", line 1, in <module>
  File "/home/morfal/.local/lib/python3.8/site-packages/tinyscript/helpers/data/transform/common.py", line 65, in _wrapper
    return f(binary, *a, **kw)
  File "/home/morfal/.local/lib/python3.8/site-packages/tinyscript/helpers/data/transform/common.py", line 104, in bin2str
    __validation(b=bs, n_b=nbits_in, n_B=nbits_out)
  File "/home/morfal/.local/lib/python3.8/site-packages/tinyscript/helpers/data/transform/common.py", line 27, in __validation
    raise ValueError(f"Bad input binary string '{v}'")
ValueError: Bad input binary string 'b'01010101''

And now, it can be fixed by ensuring that the input to ts.bin2str is of type str and not bytes :

>>> from tinyscript import *
>>> ts.bin2str(ensure_str(b"01010101"))
'U'

Ugly but I have no time right now to fix this directly into Tinyscript. I added the ensure_str invocation in the offending line on this Gist.

You can tsm update then tsm install paddinganograph --force to get the latest version.

Corrected 👍
Thank you @dhondta